AN ABSTRACT OF THE THESIS OF
Lewis Everett Queirolo
Agricultural and
in Resource Economics
Title:
for the degree of
MASTER OF SCIENCE
presented on
December 13, 1976
SUBSTITUTIONAL RELATIONSHIPS BETWEEN RAINBOWTROUT AND PANSIZE SALMON: A MARKET DEMAND
ANALYSIS
Abstract approved:
JTI-
,^ _ .
^ , - .„,— ,
/
' \
Richard S. Johnston
—f^\
—
The fundamental objective of this analysis was to isolate and
identify the factors governing the demand for domestically produced
rainbow trout in a representative west coast market, and assess the
impact on that demand, if any, of the introduction of pansize salmon.
The approach taken in this market demand study was to identify those
variables hypothesized to determine supply and demand for rainbow
trout.
Several testable hypotheses concerning the anticipated rela-
tionships were specified.
It was hypothesized that a negative rela-
tionship would exist between the price of trout at the brokerage level
and the quantity demanded at that level.
Conversely, the price of
trout at the wholesale level was hypothesized to be positively correlated with the quantity of rainbow trout demanded at the brokerage
level.
Further, the signs on the coefficients of variables accounting
for the wholesale prices of substitute items were expected to be
negative, while comparable prices at the brokerage level should be
positive, based upon the hypothesis.
It was hypothesized that the
presence of Japanese trout in the market place would have a negative
impact on the quantity of domestically produced trout demanded.
Personal disposable income was hypothesized to be positively correlated with the quantity of trout demanded.
Expectations were that
seasonal factors tend to cause trout demand to fluctuate cyclically.
An econometric simultaneous equations model was specified
from which estimates of the parameters of the demand equation were
obtained using Two Stage Least Squares techniques.
A recognition of
the limitations associated with the available data set necessitates the
emphasizing of the preliminary nature of these results.
Data on quantities and prices of rainbow trout and equivalent
price series for hypothesized substitutes were obtained through personal interviews with market participants and close observers
thereof.
The results of the study, while preliminary, tend to support the
original set of hypotheses concerning the interrelationships between
quantity of domestic trout demanded and own-price, the price of close
substitutes and seasonal demand fluctuations.
Somewhat unexpectedly,
the regression seems to have uncovered a negative income/quantity
demanded relationship for rainbow trout.
This raises some interest-
ing questions which might best be addressed in terms of hypotheses
for future analysis.
Substitutional Relationships Between
Rainbow Trout and Pansize Salmon:
A Market Demand Analysis
by
Lewis Everett Queirolo
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Master of Science
June 1977
APPROVED;
—?^*_
&—r-^
Professor of Agricultural and Resource Economics
in charge of major
Head of Department of Awyicultural and Resource
Economics
Dean of Graduate School
Date thesis is presented
Typed by Clover Redfern for
December 13, 1976
Lewis Everett Queirolo
ACKNOWLEDGMENTS
I wish to express my most sincere appreciation to several individuals without whose counsel and assistance this thesis would have
been infinitely more difficult.
In particular, I would like to acknowl-
edge Dr. Richard S. Johnston for his advice, support, and service as
major professor throughout my term of study in this department.
His
sincere concern for my well being, both academically and personally
deserve the respect and admiration I hold for him.
I should also like to acknowledge Dr. William G. Brown for his
insights and recommendations concerning econometric methods and
techniques.
His subtle wit and personal support have been greatly
appreciated.
A special thanks must be given Dr. John A. Edwards for that
seemingly always open office door and the patient counsel to be found
inside.
I wish also to express my appreciation to Dr. Roger G. Peter sen
for serving on my Committee.
It is impossible to acknowledge everyone who contributed to the
thesis which follows.
But without the support, love, and sacrifices of
my wife, Susan, and my daughters, Chelcee and Ryanne, the investment of time and effort embodied in this Masters degree would have
been quite impossible.
of us.
In a real sense this degree belongs to all four
TABLE OF CONTENTS
Chapter
I.
Page
INTRODUCTION
1
PROBLEM STATEMENT
5
III.
EVOLUTION OF THE STUDY
7
IV.
THE MARKET
II.
V.
VI.
VII.
A PRELIMINARY DEMAND ANALYSIS OF RAINBOW
TROUT IN THE LOS ANGELES MARKET
The Preliminary Economic Model for the Retail
to Consumer Level
The Preliminary Economic Model for the Wholesale
to Retail Level
The Preliminary Economic Model for the Broker
to Wholesale Level
FURTHER ANALYSIS AND HYPOTHESIS GENERATION
Subsection A: The Money Illusion, a Simultaneous
Equations Model
Subsection B: A Simultaneous Equations Model
Expressed in "Real" Price and
Income Parameters
Subsection C; A Single Equation Price Dependent
Alternative Specification
18
28
33
36
40
48
49
57
63
SUMMARY AND CONCLUSIONS
68
BIBLIOGRAPHY
72
APPENDICES
Appendix A
Appendix B
Appendix C
Appendix D
75
75
79
82
84
LIST OF FIGURES
Figure
1.
Page
Geographic distribution of Idaho food fish production
facilities.
11
2.
California standard metropolitan statistical areas.
19
3.
Los Angeles rainbow trout distribution network.
23
4.
Supply of edible and industrial fishery products,
1965-74.
29
5.
Supply of edible fishery products, 1965-74.
30
6.
Value of imports of fishery products, 1965-74.
31
SUBSTITUTIONAL RELATIONSHIPS BETWEEN
RAINBOW TROUT AND PANSIZE. SALMON;
A MARKET DEMAND ANALYSIS
I.
INTRODUCTION
The world's oceans have long provided a rich and varied source
of food protein for human consumption.
Historically, the United
States and Canada have concentrated their respective fishing efforts
and resources, at least in the North Pacific, upon a select few
species.
Among the most economically important of these have been
the anadromous salmonids, spawned and hatched out in freshwater
streams, maturing in the saltwater environment of the open ocean and
ultimately returning to complete the cycle in the freshwater streams
of their origin.
But as Brown, Larson, Johnston, and Wahle (1976)
have noted,
Increasing demands over the years for the utilization of
rivers and streams for hydroelectric power, irrigation,
flood control, navigation, waste disposal and other purposes, have presented severe hazards to the anadromous
fish. . . (pp. 1).
As a result natural runs of these several species have declined dramatically while pressures for increased resource exploitation have
mounted.
Techniques for management of this resource and alterna-
tive means of increasing depleted stocks have become high priority
issues in both countries.
But inherent in such efforts have been
numerous political and economic stumbling blocks.
As Crutchfield
and Pontecorvo (1969) noted,
The viscious and continuous political infighting that has
plagued the conservation authorities from Alaska to the
Columbia River is eloquent testimony to the participant's
awareness of economic consideration in fishery management.
The common property phenomenon and wide ranging migratory nature
of these species add an international dimension to the conservation
and management struggle surrounding natural and/or hatcheryenhanced runs.
Possible alternative sources of food fish which have been
receiving increased attention recently include aquaculture, or perhaps
more properly mariculture in the case of salmon species and saltwater species.
These range from shellfish production, either on
estuarine beds or floating rafts, to ocean ranching of salmonids,
which entails releasing fry, in the case of Pink's and Chum's, and
smolts, in the case of Coho's and Chinook's, into the sea.
They feed
and mature as would native runs of their species, returning to release
sites as part of their natural biological, reproductive cycle, where
they may then be harvested.
However, for the purposes of this study, the use of the term
"aquaculture" is intended to describe the much more restrictive fish
culture technique wherein the organism is hatched, reared, and
ultimately harvested under the constant control and management of the
production facility's staff.
In the case of salmon, the fish remain as
private property on the "farm" site throughout their life, thus
avoiding the common property dilemma facing ocean ranching operations.
In addition, theoretically, this high density impoundment cul-
ture technique can provide a continuously harvestable stock of marketable product and, of perhaps equal importance to the institutionalrestaurant trade, a uniformity of product size and quality lacking in
commercially harvested ocean runs.
A brief review of the history of salmon aquaculture might be
helpful at this point.
A Workshop on Salmonid Aquaculture; A Sum-
mary Report. , Washington Sea Grant Project (1973), indicates,
Since 1895, natural runs of Pacific Northwest salmon and
trout have been supplemented by runs of hatchery reared
fish. Because of growing demand for high quality fish in
U.S. and foreign markets, researchers have looked in
recent years to developing propagation methods such as
net-pen rearing, ocean ranching, and lagoon rearing in
attempts to find more effective and possibly more economical alternatives to traditional hatchery methods.
The results of their experiments have stimulated considerable interest from management agencies, research institutions, private investors and the publicIn reality salmonid aquaculture is not a new phenomenon. Trout
have been reared in closely managed environments for more than one
hundred years and the first "farm" in the Northwest was established
in 1909 (Klontz and King, 1975).
But in terms of anadromous
species the concept of managed food fish production is a relatively
new development.
By the late 1960's most of the technical and politi -
cal groundwork (Kane, 1970) had been laid to permit the development
of salmon aquaculture on a commercially feasible and potentially
economically viable scale.
The first production operation began in the
Puget Sound area of Washington state and by the spring of 1972 the
initial crop of pansize or mini salmon were harvested.
Since that
time the industry has grown in scope, both geographically and
methodologically, to encompass operations in Oregon, Idaho, and
Canada as well as several additional production sites in Washington
state.
Investors have ranged from large national and international
corporations such as Union Carbide, Ralston-Purina, and Weyerhauser to fishermen's cooperatives and several consolidated Indian
tribes (Deloria, 1975).
The rearing techniques have been as numer-
ous and varied as the companies in the industry.
Presently several fundamentally important questions concerning
the economic feasibility of the pansize salmon product remain
unanswered-
Thus the ultimate future of this phase of the salmon
aquaculture industry is still unresolved.
However, it is abundantly
clear that resolution of these issues is of critical importance both
for the members of the industry and the salmon fisheries in general.
II.
PROBLEM STATEMENT
Aquaculture, the process of rearing fish and shellfish in high
density, closely managed habitats for purposes of human consumption,
has been likened to the feed lot concept of livestock production
(Lannan, 1975).
But unlike its agricultural counterpart, aquaculture,
and for that matter fisheries products generally, have received very
little attention in the way of objective quantitative demand analysis, as
discussed in the introduction.
Legislative and administrative deci-
sions which may have long range repercussions upon the food fish
industry have often been made without a full appreciation for the
economic impact they entail.
In most cases this has been due pri-
marily to the fact that the information necessary for informed decisions simply was not available.
Knowledge of the nature of demand for a product would be
invaluable, not only in terms of allocating scarce resources efficiently on the part of public decision makers, but equally as well to
members of the industry and potential entrants, in the process of
making sound investment, management, and production decisions.
Intense production of aquatic organisms on a scale capable of
providing sufficient stocks of marketable product requires substantial
capital investment and, by its nature, faces very high production
costs (Richards, Mahnken, Tanonaka, 1972).
An understanding of the
dimensions of current and potential demand and the substitutional
relationships governing market performance would provide insight
into the effects on total revenue associated with various price and
output levels for aquaculturally produced food fish products.
III.
EVOLUTION OF THE STUDY
Some preliminary work on the economic and market potential
of the so called, "pansize" or "mini" salmon had generated a great
deal of interest and guarded enthusiasm among those in and around
the aquaculture industry.
Most of this early research centered upon
the technology and cost of production associated with salmon aquaculture.
However, some work on product marketability was undertaken.
Test marketing of the "single serving" salmon product took place in
the east and midwest as well as in California.
The character of most
of this market work emphasized a non-quantitative "opinion poll"
format, wherein the new product was evaluated on the basis of several
criteria by brokers-wholesalers, institutional buyers, and restaurant
managers and chefs.
The results indicated a very rosy future for
pansize salmon, but astute observers noted the danger in relying too
heavily upon these preliminary results.
As the Evaluation of the
Commercial Feasibility of Salmon Aquaculture in the Puget Sound,
undertaken by Jack A. Richards, Conrad V. M. Mahnken and George
K. Tanonaka of NMFS (1972) pointed out, "There is no established
market for this new, unique product--'plate size' salmon" (p. 13).
As a result no hard data on performance of the product in the market
place were available upon which to make sound public and private
policy and investment decisions concerning such issues as licensing.
8
enhancement programs, water use, discharge levels of waste and
investment in appropriate monitoring programs, etc
The need for
quantitative economic analysis of potential consumer demand for the
product was clear.
As pansize salmon received wider exposure in
the market place the data would, it was assumed, be forthcoming.
As mentioned earlier, the initial commercial pilot project
designed to rear "mini" salmon was undertaken by Ocean Systems,
Inc. in the Puget Sound, Washington.
The first harvest of marketable
product was expected in the spring of 1972.
Since that time several
other production facilities have been constructed, both in the
Puget Sound area and elsewhere in Washington State and Oregon.
Thus by July of 1975, when this study was in its formative stages,
the expectation was that sufficient time series and/or cross-sectional
data to do a quantitative demand study would be available.
Because of the reasonably close proximity of Oregon Aqua Foods
Inc. , of Newport, Oregon, to Oregon State University and the close
working relationship of "Ore -Aqua's " founder, Dr. John R. "Jack"
Donaldson, with this institution, contact, exploring the possibility of
obtaining cooperation and project input from the aforementioned firm,
was made.
Mr. Robert Cantrell, business manager of Oregon Aqua
Foods, was both interested and most cooperative.
However, after
several visits to the company's Newport facility it became apparent
that the available production, price, and shipments data were
insufficient to support a meaningful statistical analysis.
In the course of developing testable hypotheses for the pansize
salmon study, it became apparent that rainbow trout was perceived
by pansize salmon producers to be one of the strongest potential substitutes for the new product in the market place.
Because data
limitations prevented direct estimation of the demand for pansize
salmon, it was felt that using rainbow trout, a widely distributed,
well established aquaeulturally produced food fish, as a close proxy
for pansize salmon might permit the observation and identification of
the forces governing the performance of both products in the market
place.
It was under this assumption that the focus of this study
shifted to an estimation of the demand for rainbow trout and the
impact, if any, which the arrival of pansize salmon in the market had
on that demand.
Because trout has been aquaculturally produced and marketed
in the U. S. for more than 106 years (Klontz and King, 1975), it was
assumed that copious amounts of data would be available and that some
demand analyses had probably already been done.
However, after an
exhaustive search it became clear that neither assumption was correct.
Thus the current study is a pioneering effort in the estimation
of the demand for rainbow trout.
It is also somewhat unique in
attempting to explore the impact of a "new" product on the demand for
an "established" one.
10
An indepth historical survey of the trout aquaculture industry
is beyond the scope of this thesis.
An excellent compendium of
aquaculture in the United States is available for interested readers in,
Aquaculture in Idaho and Nationwide, Idaho Department of Water
Resources, Klontz, and King, 1975.
For clarity sake however, a
few highlights of the evolution and character of that industry must be
made.
The state of Idaho is responsible for 90% of the nation's total
production of rainbow trout.
Estimated total trout output, in Idaho,
by 1980 is expected to reach 39 million pounds and contribute nearly
$50 million to the state's economy (Klontz and King, 1975).
The
industry itself is comprised of a relatively few trout producing companies, most of which are located within a ten mile radius of Buhl,
Idaho (see Figure 1).
Among the some 14 companies and 28 produc-
tion facilities, the industry is dominated by three major integrated
operations accounting for 80% of the total output.
The potential for cooperative marketing and promotional
activities appear obvious.
Nonetheless, every attempt at such
coordination has resulted in varying degrees of failure.
The industry
might best be described in terms of intense personal rivalries bordering on physical violence in some cases.
Even in the best of circum-
stances, highly price-conapetitive behavior has characterized this
industry throughout the post World War II period.
Demand analysis
11
DISTRIBUTION OF FOOD AND GAME FISH HATCHERIES
IN IDAHO
LEGEND
HATCHERIES
■
COMMERCIAL
▲
FEDERAL
•
STATE
N
Figure 1.
Geographic distribution of Idaho food fish production
facilities. Source: Aquaculture in Idaho and Nationwide,
1975, Idaho Department of Water Resources.
12
and market research has, for the most part, been lacking as a result
of apprehension on the part of producers to provide accurate detailed
production data.
Klontz and King (1975) noted in their recent study,
The food animal industry unlike hard-goods industries,
must move their product when it is a certain size or it
becomes a financial liability--it keeps on growing and
must be fed to keep it presentable for the consumer.
Current (marketing) practices are inadequate to promote
increased sales of food fish over the counter or in restaurants. The primary problem with marketing (in Idaho trout
aquaculture) is that it has been guided by production rather
than the other way around (p. 69).
Sensitive to the increasing awareness on the part of industry
members of the need for market analysis and yet conscious of their
historical behavior, project members sought initial contacts within or
around the Idaho food fish industry.
The first and surely the most
cooperative source of ongoing information concerning the industry in
general, was Dr. G. W. Klontz of the University of Idaho.
In addition
to providing a wealth of background information, Dr. Klontz was
instrumental in identifying and, in several instances, actually introducing key individuals within the industry.
With his assistance an
onsight visit was arranged with one of the major producers for early
January, 1976.
However, a number of unforeseen events prevented
the scheduled meeting from taking place.
It rapidly became clear that
the data being sought would not be available from the primary sources,
i.e. , the major Idaho producers, at this time.
13
Subsequently, contact was established with Mr. Robert Erkins,
who had been instrumental in the growth and development of Thousand
Springs Trout Farm, Inc. into one of the three most significant trout
operations in Idaho.
Currently, Mr. Erkins co-authors the fisheries
industry publication, "The Gruber-Erkins Newsletter,
longer directly affiliated with Thousand Springs.
and experience proved to be invaluable.
n
and is no
However, his insight
After an extensive conversa-
tion at his home in Bliss, Idaho, during which the project was discussed in some detail, Mr. Erkins suggested that the most likely
source of available data on price and quantities would be through the
brokerage network within the trout marketing system.
It was hoped
that brokerage firms, being somewhat closer to the process of
marketing a product, could be persuaded to provide the aggregate
shipments data and price lists necessary to successfully carry out the
analysis. Trout, while a valuable food product, actually accounts for
a relatively small percentage of a brokers gross sales volume, and
for this reason relinquishing sales data would be less apt to evoke
concern.
Although it was made clear from the start that all results
of this study would be made public, it was also pointed out that
specific data would be held in strictest confidence and only marketwide
aggregate figures would be incorporated within the published results.
The exact location for the demand study was, at this time, still
an unresolved issue.
It seemed clear that, for the purposes of this
14
study, a market should be identified in which trout and its closest
hypothesized substitutes, namely commercially harvested ocean
salmon and aquaculturally produced pansize salmon, were most consistently in close competition for the consumer's dollar-
Because of
the limited distribution of the pansize salmon product and resource
constraints associated with the research project, a west coast outlet
seemed most desirable.
Seattle and Portland were rejected on sev-
eral counts, perhaps chief among them being the availability of large
quantities of trout and salmon readily accessible to sport fishermen
from the two metropolitan areas.
Erkins indicated that both Seattle
and Portland were traditionally "poor" trout markets in terms of consistent sales volume.
San Francisco had seemed a highly promising
market based upon its worldwide reputation as a seafood capitol and
the fact that pansize salmon had been introduced there early in the
test marketing stage.
ing San Francisco.
But despite these facts Erkins suggested avoid-
Apparently, in part because of its reputation and
the substantial tourist trade, San Francisco is frequently a "dumping
ground" for excess stocks and for this reason is not perceived to be
a representative market.
Los Angeles, on the other hand, seemed
free of the possible biases associated with the Seattle, Portland, or
San Francisco markets and additionally represented a substantially
larger potential consuming population than did any one of the other
three.
It was Mr. Erkings' impression that Los Angeles also
15
represented the market in which ocean run salmon, pansize salmon,
and rainbow trout had been most consistently available over the longest continuous series of years.
This conclusion was later verbally
supported by several other sources, including those who were currently producing pansize salmon.
This information suggests that the
Los Angeles market might best serve as the target area for the project.
Bear in mind that the initial purpose of this study centered on
pansize salmon.
As the preceding section has attempted to demon-
strate, this goal could be satisfied only indirectly.
Thus, the prin-
cipal objective of this research became an effort to isolate and identify
the forces governing the demand for rainbow trout and the impact upon
that demand of changes in the price of several potential substitues.
Of obvious interest was the impact on the demand for trout of the
appearance of pansize salmon in the market.
Most specifically, the study intends to (1) develop a model of
the structure of the Los Angeles market which identifies the factors
hypothesized to influence the demand for rainbow trout,
(2) estimate
the magnitudes of own-price, cross-price, and income elasticities
associated with rainbow trout in this nnarket and (3) assess the potential impact of various pricing and marketing strategies upon total
revenue of brokers and ultimately producers, of rainbow trout.
16
As mentioned earlier, very little has been done in terms of
identifying criteria and evaluating techniques for assessing the impact
of a "new" product in a highly competitive market situation.
A sig-
nificant objective of this study, then is to seek to establish an
analytical framework within which a unique new product, namely pansize salmon, might be quantitatively evaluated in terms of market
impact and potential, despite the lack of substantial data on historical
or cross - sectional market performance.
Preliminary contacts were made with brokerage firms in the
Los Angeles market serving as representatives of the major Idaho
trout producers on the basis of a list of distributors provided by Mr.
Erkins.
Dr. William Jensen, who has been a participant in and a close
observer of the Pacific coast fishing industry, was instrumental in
establishing contact between industry representatives and project
members.
Despite the efforts by Mr. Erkins and Dr. Jensen to lay the
appropriate introductory groundwork, the project's reception was
mixed.
While some data were immediately made available, several
of the most significant brokerage houses either flatly refused to
cooperate or required express permission from their Idaho producers
prior to releasing data.
circle.
Thus the study seemed to have come full
Despite this resistance and armed with at least some hard
data and the impression that a significant portion of the market was
17
therein represented, analysis began.
It quickly became apparent however that substantial portions of
the market were not being accounted for within the available data and a
second round of information search was undertaken.
Subsequently
two of the principal suppliers of the Los Angeles market, namely
Clear Springs Trout Company and Thousand Springs Trout Farm, Inc.,
had agreed to release data on their shipments into southern California.
At this point it was believed that 80-90% of the trout entering L.A.
were being accounted for by the data acquired in L.A. and that just
promised.
Unfortunately, before the data from Clear Springs Trout
Company could be obtained a fire destroyed one of the buildings at the
company's Idaho production and processing facility and all sales
records were lost.
While Clear Springs is currently the world's largest trout
producer, their participation in the Los Angeles market has been very
recent, spanning only two to three years.
Thus, it was felt that the
loss of data on that portion of the market, while a most regrettable
incident, would not significantly hamper the analysis.
Further, data
which represent the most recent market participation on the part of
Clear Springs Trout Company have been provided by Mr- Larry Cope,
general manager of the Idaho operation, and could be utilized for
validation of the model.
18
IV.
THE MARKET
The market distributional network referred to in this analysis
is geographically identical to the standard metropolitan statistical
areas, (SMSA's), of Los Angeles, Orange, Ventura, Riverside, and
San Bernadino counties, California, as set forth by the United States
Bureau of Census (see Figure 2).
This specific group of counties in
the Southern California basin is also referred to by the Bureau of
Statistics as the Los Angeles standard consolidated statistical area,
(SCSA), the dark shaded area in Figure 2, and may be referred to by
either title within the body of this study.
It was noted in the course
of data preparation that delineations of SMSA's have tended to
fluctuate over time and therefore appropriate care has been taken to
assure consistency in the statistics incorporated in this anslysis.
Through interviews with representatives of several of the major
brokerage firms in the Los Angeles area it became apparent that this
combination of standardized areas described closely, and most satisfactorily, the boundaries within which trout arriving in L. A. were
distributed.
Brokers in San Luis Obispo area serve territories to the
north and San Diego distributors are significant agents in the extreme
south.
Some overlapping does occur at the margins, and in one case
an L. A. broker reportedly was handling the Las Vegas trade for the
trout farm his firm represents, but, generally speaking, the vast
19
*.....f,i
l\
.
~Sl^
T^TTT]Avx^
V_ !
\^i --^
) / PFOi- Wd
Vjk't^^SR.
N
Vallejo - Fairfield - NapS^^^Jg^gLJ^y
*v
San Francisco - ^nJ^^Tf^ ■■••7 ^S-
/?/\
W&lPS*w....... ^<• jf1; - K
Santa Cruz-
iFresno
Salinas - Seaside - Monterey-
Bakersfield
.. * * »'
Santa Barbara - Santa Maria - Lompoc^imm^ll
IS^^^ernardmo - ""^
Oxnard - Simi Valley - VenturaLos Angeles - Long Beach Anaheim - Santa Ana - Garden Grove'
.San Diego
CALIFORNIA
Standard Metropolitan Statistical Areas
as of July 1,1915
Figure 2.
California standard metropolitan statistical areas.
Source: California Statistical Abstract, 1975.
20
majority, perhaps greater than 95%, of the rainbow trout entering
through the Los Angeles based brokerage houses are distributed and
consumed within the geographic boundaries described above.
The dimensions of the Los Angeles SCSA, in terms of a potential consuming population, are substantial as the population and
personal disposable income figures in Table 1 denote.
However,
indications are that a consistent marketing program or strategy for
exploitation and development of this potential market has been lacking
historically-
Industry sources explained that, traditionally, as the
demand for rainbow trout increased in the Colorado resort market,
incidently the largest single trout market in the U.S. during the heavy
tourist season, the producers had characteristically pulled their product out of Los Angeles, and other lesser markets, in order to
accommodate the Colorado trade.
This, obviously, disrupted sup-
plies of the product and alienated customers, thereby necessitating a
complete rebuilding of the L. A. distribution network for rainbow trout
each year.
The phenomenon continued until the early to mid ^GO's
when some of the Idaho producers began to recognize the inherent
costs and inefficiencies of such a supply strategy and ceased the
practice.
Additional production capacity resulting in increased sup-
plies of rainbow trout also contributed to the elimination of this
historical pattern.
Table 1.
Estimated total population of selected California counties 1970-75 and adjusted gross and
median income for selected California counties (1975).
County
July 1,1970
July 1,1971
July 1,1972
July 1,1973
July 1,1974
July 1,1975
6,955,500
1,653,500
514,200
694,600
427,000
6,970,000
1,694,900
526,600
698,300
438,200
Estimated total population of selected California counties 1970-75:
Los Angeles
Orange
Riverside
San Bernadino
Ventura
7,047,100
1,431,600
461,400
685,200
381,400
7,071,200
1,471,000
474,000
689,500
389,800
Gross -($1,000)
6,988,900
1,526,700
488,500
690,500
404,200
6,966,200
1,592,300
501,600
691,400
415,200
Median
Adjusted gross and median income for selected California counties (1975):
Los Angeles
Orange
Riverside
San Bernadino
Ventura
$28,735,988
7,202,367
1,584,552
2,146,000
1,571,157
$ 9,121
10,684
8,544
9,364
10,605
Source: California Statistical Abstract, California Department of Finance, 1975; pp. 7.
22
Rainbow trout enter the Los Angeles market in several forms
including, fresh or fresh-frozen, dressed or boned.
However, data
limitations preclude estimating separate demand relationships for
each.
Thus this study will, by necessity, aggregate these groups and
estimate the demand for rainbow trout in total, ignoring the particular
form the product takes.
Interpretation of the results, thus, will
require one to be sensitive to this limitation.
The single exception to
this format will be in the case of trout categorized as "large".
As
has been explained in the preceding chapter, once trout reach optimal
market size they must be harvested or very soon they become too
large to command a premium price in the market place.
Historically
this has occurred periodically and these "large" trout have been
offered for sale at greatly reduced prices, usually in significantly
larger shipments than are characteristic for the "prime" product.
For the purposes of this demand analysis these "large" trout are
hypothesized to be highly price competitive substitutes for the smaller
sized fish.
When referring to the premium product the term "rainbow
trout" will be used.
"Large" trout will be so identified henceforth.
The current Los Angeles distribution network, as it relates to
rainbow trout, might be represented by a flow chart such as the one
in Figure 3.
Notice, with the exception of imported Japanese trout,
all the trout entering the Los Angeles market are the product of the
"far west" or "mountain" states, these being dominated far and away
ALT£#A/ATl\f£
Wfsrccvter
EAST COAST
^A"RJC£T'S
Figure 3.
Los Angeles rainbow trout distribution network.
24
by the Idaho "farms".
While there are trout production facilities in
several eastern states, most notably North Carolina and Pennsylvania,
indications are that none of their product reaches the Los Angeles
market.
There are rainbow trout production facilities in the state of
California as well.
However, there is no competition between these
trout and the Idaho product.
The California fish are ultimately
destined for fee fishout ponds, not the retail consumption market,
according to California Department of Fish and Game representatives.
Thus, as the flow chart indicates, the rainbow trout are
marketed by brokerage representatives for the producers.
Histori-
cally, each of the "farms", has secured a single broker in the Los
Angeles market to handle the distribution of its product.
These
relationships, between producer and broker, have been subject to
For the sake of accuracy it should be noted that rainbow trout
are produced in relatively significant numbers in Montana, Colorado,
and Wyoming. Several other western states have production facilities
but do not produce sufficient quantities of product to merit attention.
But the volume of output of all of the above mentioned producers is so
minor when compared to the Idaho industry's production that for purposes of this analysis the term "Idaho rainbow trout" will be used
interchangeably with the term "domestically produced rainbow trout".
Conservative estimates of Idaho's dominance of the industry put the
portion of total national production of rainbow trout reared and harvested in Idaho at 90-95% (Klontz and King, 1975). The proportion of
domestically produced rainbow trout distributed in the L. A. market
originating in Idaho is even higher. Therefore, no attempt will be
made to differentiate between Idaho, Montana, Colorado, and Wyoming
trout entering L. A. The volume of non-Idahoan, but domestically
produced trout has historically been very insignificant when they
appeared at all, at least over the period observed in this analysis.
25
fluctuation or turnover over time.
However, characteristically,
there appears to have been a "one to one" relationship between producer and broker, at any point in time.
As the Los Angeles rainbow trout market has been dominated,
historically, by primarily three to five Idaho trout companies, so too
the number of brokerage firms handling the product have been dominated by three to five significant brokerage houses.
Industry esti-
mates place the portion of the total L.A. market served by the three
principal "farms", and therefore the three principal brokers, at
85-90%, with the remaining 10-15% divided among half-a-dozen other
producers.
These percentages have fluctuated somewhat over time;
however, they are essentially indicative of the relative influences in
the L. A. market until very recently.
Again referring to the flow chart, the small number of
brokerage houses, acting as exclusive representatives of their particular producer, service a substantially more numerous intermediate
or wholesale level of the market-
Purchases by these wholesalers,
historically, have tended to fluctuate between brokers, i. e. , between
"farms", as the availability of a particular producer's product varies.
However, again, at any one time a wholesaler probably only distributes a single "farm's" productAt the next level are the final consumption purveyors.
As will
be explained in more detail in the following section, this level of the
26
domestic rainbow trout distribution network is primarily an
institution-restaurant trade.
The number of these establishments,
either serving rainbow trout or as potential markets for the product,
is staggering in a megalopolis the size and character of the Los
Angeles SCSA.
Historically, these retailers have born the brunt of
the supply fluctuations described above in terms of relatively highest
costs incurred as a result of the loss of supply.
And, perhaps as a
result of this, indications are that, at least some of these establishments, maintain an interlocking supply relationship with several
wholesalers serving different Idaho producers.
This may cease to be
the case as supplies become more stable but over the period of this
analysis some of this behavior was reported to be persisting.
The majority of the over-the-counter retail trade has traditionally been met by imported trout, primarily from Japan.
However,
some few domestic rainbow trout do pass through the primary distribution network and arrive at the market outlet purveyor.
Addi-
tional Idaho rainbow trout bypass the broker-wholesaler-retailer network and are purchased directly from Idaho producers by "food
chain" buyers for ultimate distribution to their local retail outlets.
Safeway is reportedly the prime example of this behavior in the L. A.
market area.
However, the volume of domestically produced trout
which arrive in the over the-counter market is quite small, at present; perhaps no more than 5% of the total volume of rainbow trout
27
entering Los Angeles.
Japanese trout, as was mentioned, account for
the remaining volume and may be distributed through either of these
two alternative systems.
Beyond the L. A. market, as detailed above, Idaho rainbow trout
are marketed throughout the United States and there are tentative
plans to investigate the potential of export markets, especially
Europe.
Thus these external markets constitute important potential
shifters in the supply of rainbow trout available in the Los Angeles
market.
This will become more obviously significant in terms of the
economic model which will be specified in the following section.
28
V.
A PRELIMINARY DEMAND ANALYSIS OF RAINBOW"
TROUT IN THE LOS ANGELES MARKET
Per capita consumption of edible food fish products has
remained approximately constant over the post W. W. II years.
Most
available sources place the figure at between 10 and 11 pounds per
capita annually.
"But whether per capita consumption increases or
not, an increased demand for food fish, as for all foods, must come
in the future as the population explosion continues" (Bell, Hazelton,
p. 23).
If projections are correct, the population of the United States
will soon reach the 225 million mark.
The impact upon total con-
sumption, assuming constant per capita consumption, will result in
an additional 385 million pounds of edible fish being required just to
accommodate this growth (Bell and Hazelton, 1967).
At present a majority of the fish products, both edible and
industrial, being consumed in the United States are imported.
Gen-
erally this trend has been on the increase in recent years as indicated by Figures 4 and 5.
Perhaps more dramatically, the market
value of these imported fish products has more than doubled over the
last decade (see Figure 6).
Batie (1974) noted "It is not immediately
clear -why this should be so.
The United States is one of the largest
seafood markets in the world, and U.S. fishermen have an obvious
geographic advantage in meeting this demand. " Imports have characteristically enjoyed a price advantage relative to their U.S.
Billion pounds
Commercial landings
1965
1966
Figure 4.
19^7"
1968
19*69
1970
1971
1972
1973
1974
Supply of edible and industrial fishery products, 1965-74. Source:
Current Fisheries Statistics No- 6700. Fisheries of the United
States, 1974. March 1975, U.S. Department of Commerce
National Oceanic and Atmospheric Administration National Marine
Fisheries Service.
Billion pounds
Total supply
Imports
Commercial landings
197^
Figure 5.
Supply of edible fishery products, 1965-74. Source: Current
Fisheries Statistics No. 6700. Fisheries of the United States,
1974. March 1975, U.S. Department of Commerce National
Oceanic and Atmospheric Administration National Marine
Fisheries Service.
o
Billion dollars
1600
1200
800
Edible
400
1965"
19^
Figure 6.
1967
1968
1969
1970
1971
1972
1973
197*
Value of imports of fishery products, 1965-74. Source: Current
Fisheries Statistics No- 6700. Fisheries of the United States,
1974. March 1975, U.S. Department of Commerce National
Oceanic and Atmospheric Administration National Marine
Fisheries Service.
32
counterparts.
However, at least in the case of rainbow trout, the
quality of the imported product has been somewhat lower.
As a
result indications are that the two trout products, domestic rainbow
trout and imported trout, have traditionally found their way into different consumptive markets.
As noted in a preceding section, while
some domestically produced rainbow trout ultimately arrive in the
meat display or frozen food counter of retail markets, indications are
that the overwhelming majority of the "prime" quality domestic product reaches the consumer through the restaurant-institutional
system,
2
while imported trout generally account for the bulk of the
over-the-counter trade.
Very recently, U.S. importation of foreign
produced trout, historically dominated by the Japanese but also including significant shipments from Denmark, have been declining as a
result of increased demand for trout within Japan and some European
countries, and because of disease problems in Danish stocks
(Robinson, 1972).
It would therefore appear reasonable to assume
that some future market adjustment in the domestic product's distribution may be inevitable.
2
It has been observed and varified through industry sources that
upwards of 90-95% of all domestically produced rainbow trout are consumed within the restaurant-institutional trade. Therefore references
made above to "retailers" are intended to identify restaurants, institutions, and their suppliers, at the level immediately below wholesale, when appropriate "market outlets" refer to retail food stores
such as grocery or seafood outlets.
33
In an effort to isolate and identify the factors governing the
movement of rainbow trout within and between the several levels of
the Los Angeles distribution network, a preliminary economic model
has been hypothesized.
If one accepts the existence of consumer utility functions and
further accepts that consumers act to maximize their own utility subject to budget constraints, then certain consumer demand behavior
can be anticipated, at least to the extent that they constitute testable
hypotheses.
The relationships responsible for the retail performance
of rainbow trout in the L.A. market, based upon the above assumption, can be expressed in terms of the functional interaction between
the quantity of trout in the market, its own price, the prices of close
substitutes and/or complements and the personal disposable income
of the potential consuming public.
The hypothesized specification of
the retail segment of the economic model can be expressed, using the
notation in Foote (1958, p. 8) as:
The Preliminary Economic Model for the Retail to Consumer Level
Cts
d
pCt
r
CP88
CSvb
r
r
or alternatively in per capita terms, where
Z
"N"
3
4
stands for popula-
tions
Qct
-£-: PCt,PCpSS,PCSvb,Y/N,D?,DvD4
N
r
r
r
Z
3 4
34
and assuming the identity,
QCt = Qrt = Qt0tal
d
s
where.
ct
ct
Q, or Q /N = (per capita) quantity of rainbow trout demanded
d
d
by L. A. consumers at the final consumption level
P
P
P
ct
= price of rainbow trout from retailer to consumer
r
CDS S
= price of pansize salmon from retailer to consumer
cs vb
= price of other food protein substitutes from retailer
r
to consumer
Y or Y/N = (per capita) personal disposable income in the Los
Angeles SCSA
D^.D-.D. = binary variables included to account for seasonal
shifts in consumer demand on a quarterly basis.
All variables, excluding
D?, D,,
and
D ,
have time subscripts
where the time interval is one month.
Based upon the assumptions and hypothesized relationships
stated above it is expected that, by specifying quantity demanded as
the endogenous or dependent variable and the other variables in the
equation as exogenously determined, the sign of the coefficient on the
"own-price" variable,
P
ct
,
will be negative.
By including "own-
price" of rainbow trout as a predetermined variable in the model, the
35
assumption is made that forces beyond the scope of the L. A. market
work to deternainethe price in L. A.
For example, Los Angeles, while
a significant market, is only one of several market areas into which
producers can ship trout.
To some extent then the price in L.A.
might be assumed to be dependent upon the price/quantity relationships of these other markets.
price of substitutes, including
The signs of the coefficients for the
P
cos s
r
and
P
c s vb
,
r
would be
expected to be positive, as would the sign of coefficients on personal
disposable income.
The second distributional level accounts for the wholesale to
retail movement of rainbow trout in Los Angeles.
As in the preced-
ing equation the quantity of rainbow trout, total or per capita, is
treated as an endogenous variable and is specified to be a linear
function of the price of rainbow trout paid to the wholesaler, the price
of rainbow trout received by the retailer, the price of substitute food
proteins confronting the retailer as well as the prices the retailer
could command for these same substitute items, and some variable
accounting for retail costs associated with including rainbow trout on
the menu, e. g. , menu inserts, table tents or other specific promotional costs.
In addition, because some product does move into
"market outlets" a variable accounting for associated "market outlet"
costs has been included.
second stage becomes:
Thus the economic specification of this
36
The Preliminary Economic Model for the Wholesale
to Retail Level
Q'*: Prt,PCt ,PrSvb,PCSvb,I ,C
d
W
r
t
t rt-l Wt
t-l * '
where:
Q
rt
rt
or Q. /N = (per capita) quantity of rainbow trout demanded
w,
w,
d
d
by retailers at wholesale
rt
- cprice of rainbow trout to retailer at wholesale
w
ct
P - price received for rainbow trout by retailer
P
P
r s vb
= price of substitute seafood items to retailer at
w
wholesale
P
cs vb
= price received for substitute seafood items by retailer
r
I = some index of costs associated with featuring rainbow
trout as a menu item
C = costs associated with maintaining fresh and fresh
frozen fish in a meat display case of an over-thecounter retail outlet.
Note the inclusion of the subscript
economic model.
As before,
variables subscripted
t-1
t
t
for each variable in the
equals the one month interval; thus
are lagged by one month.
for this decision is discussed on pages 38 and 39-
The rationale
37
The third level of distribution in the L. A. market is that of
broker to wholesaler.
It is this stage of the marketing network which
will be most closely scrutinized in this analysis.
This decision was
made primarily on the basis of two criteria, the first being the need
to limit the scope of the research to accommodate time and resource
constraints.
The second, and perhaps more utilitarian reason, was
some initial indication of availability of cooperation from Los Angeles
brokers and wholesalers and the subsequent contribution of primary
data on this facet of the distribution network.
However, as discussed
in the following section, data problems surfaced even at this level,
requiring the adoption of an internaediate strategy to solicit much
needed additional cooperation from industry sources.
As has been alluded to above, the initial series of efforts at
data acquisition were not wholly successful.
It became apparent,
rather quickly, that significant portions of the total quantity of rainbow trout entering the Los Angeles market were not being accounted
for by the data in hand-
Confirmation of this suspicion came when,
in a subsequent phone conversation with Mr. Erkins, he noted the
sources that had provided data constituted perhaps no more than 35%
of the total product movement in this market during the period 1964
through 1975, under consideration.
It was decided that the data currently being withheld might be
obtained if those in possession of the missing information had a
38
clearer understanding of what the analysis intended to do and how
critical to the study's success their participation was.
As a means to
this end a single equation model was specified and coefficients estimated using ordinary least squares technique, and utilizing a much
shorter time series for which approximately 90% of the market was
represented by the data, namely May, 1972, through December,
1975, in hopes that the results might be used as a descriptive tool in
an effort to secure wider cooperation on the part of industry members.
It was recognized that the results of this model's analysis would,
very likely, be unsatisfactory in terms of describing the actual
behavior within the market, as a result of the incomplete data set and
brevity of the time series.
However, as a first approximation, given
the obvious problems just cited, and recalling that the principal reason for this first stage of the analysis was to demonstrate the general
thrust of the investigation, the time and energy invested seemed
completely justifiedThe economic model itself is described below.
The equation
specifies quantity of rainbow trout demanded by Los Angeles' wholesalers as the dependent variable.
to be pre-determined.
All other variables are assumed
Price paid and received by wholesalers for
rainbow trout and prices of hypothesized substitute food proteins, are
deflated by the WPI (wholesale price index) for "Farm products,
processed foods and feeds, " U.S. Bureau of Labor Statistics.
In
39
addition, prices at the wholesaler to retailer level are lagged by one
time period in order to assess the validity of the following hypothesis.
The analysis is seeking to identify the derived demand curve for the
wholesale segment in the marketing structure.
It is reasonable to
assume, it is hypothesized, that planning decisions, at the wholesale
level, are made in light of the most recent price information available.
In essence, the wholesaler is demanding trout as an "input
factor" to his final product.
Whether he processes, repacks, or
simply distributes the product, the analysis is fundamentally the
same.
Thus the wholesaler is confronted with several sets of prices
when making decisions concerning trout.
They include the cost of
rainbow trout he must pay to obtain the "input", the price he is able
to command for rainbow trout from retailers and the prices and costs
associated with substitute inputs, i. e. , alternate seafood products.
This is the thought process which led to the decision to incorporate a
lag in the price variables of this equation.
An additional benefit
associated with this "first approximation" process was an opportunity
to scrutinize the resulting output in order to identify additional
hypotheses which might then be more fully explored using the more
complete data set.
40
The Preliminary Economic Model for the Broker
to Wholesale Level
The economic model specified was as follows;
Qwt:
d
pwtj p
b
rt
w
prmSj
w
pWirgt^ prpSSj ptrn
b
w
^
D3>D4, Y,D
and assunrxes the identity:
Qwt =
d
wt
s
= Qtotal
recalling the colon in read "depends on", a comma is read "and",
Q
wt
- quantity of rainbow trout supplied by producers to the
L. A. market
Q
wt
= quantity of rainbow trout demanded by L. A. wholesalers
at brokerage level
wt
P, = price of rainbow trout from broker to wholesaler
P
rt
- price of rainbow trout from wholesaler to retailer
w
Pw
P
rm s
= price of medium salmon from wholesaler to retailer
w
Wj?r 21
P,
P
= price of rainbow trout from wholesaler to market outlets
TL ID S S
w
= price of "large" trout from broker to wholesaler
= price of pansize salmon from wholesaler to retailer
41
Y = personal disposable income in the L.A. (SCSA); not
deflated
D
=
binary variable indicating presence or absence of
imported trout in market
D-, D_, D
= binary variables included to account for seasonality
shifters on quarterly basis.
All variables have "time" subscripts, where the time interval is one
month, except
both
wt
P,
b
and
D,,, D_ and
P
rt
w
D..
The perceived need to include
in the equation has been addressed above,
The remaining variables in this model and for that matter, subsequent specifications, were included on a variety of grounds.
The price of rainbow trout at the market outlet level was
included in this preliminary model on the assumption that
P
w
would be accounting for shifts in demand resulting from changes in
consumer demand for rainbow trout over the counter.
could potentially cause a significant change in
Q
wt
d
This shift
and necessitate
price and allocation adjustments throughout the market.
Through discussions with seafood wholesalers in L.A. it was
determined that "medium salmon" probably served as one of the
closest substitutes for rainbow trout in this market.
This hypothesis
was implicitly incorporated into the model by the inclusion of
P
rms
w
42
Once again, in order to account for a hypothesized substitutional relationship, this time at the same broker to wholesaler level,
the variable
wirgt
P,
was included in the model.
Large trout appear
in the market place periodically and usually in very large volume.
Thus, if these two trout products are substitutes for one another, a
shift in demand for rainbow trout should be detectable in conjunction
with the flood of large trout onto the market place.
The inclusion of the price of pansize salmon has been discussed
in some detail already.
Suffice it to say, pansize salmon in terms of
its relationship to rainbow trout is a very important issue in this
analys is.
The inclusion of personal disposable income in a wholesale
demand model is intended to account for the very significant influence
retail demand has on the derived demand at wholesale.
Y also tends
to reflect the general economic condition of the potential consuming
public which is targeted to be the final consumptive market for this
product.
D
appears as a binary variable in the model to reflect the
potential influence of Japanese trout, as a demand shifter, upon total
quantities of domestic rainbow trout moving through the Los Angeles
distribution network.
Finally, a series of three dummy or binary variables are
included to detect any significant seasonal shift in demand on a
43
quarterly basis.
Several preceding studies on seafood demand have
tentatively identified seasonal demand fluctuations owing to a variety
of factors including religious holidays, traditional non-seafood main
course dishes, i. e. , Thanksgiving turkey or Christmas ham, etc
Industry members are divided as to the significance and timing of such
seasonal shifts in demand for rainbow trout in the L.A. market.
Thus, the need to attempt to isolate the seasonality factor appears
justifiable and desirable.
,.....,
.
.
Implicit in the identity equation
^wt ^wt ^total
Q
= Q
- Q
is the
assumption that no stocks of rainbow trout are held at any of the
market levels in L.A. or that they are constant over time.
The econometr ic model, estimated coefficients, t statistics,
and elasticity values can be found in Appendix A.
The most obvious result of the O. L. S. estimation was that only
slightly more than half of the variance associated with the dependent
variable, 56%, was being explained by the variables in the model,
based upon the observed data available.
This would seem to reaffirm
the contention that the data being withheld would be of significant
importance to a complete evaluation of the problem.
It is, however
encouraging to observe that the signs on all coefficients are consistent
with hypothesized relationships between the endogenous variable and
each of the predetermined variables.
44
The entering F value for the complete regression indicated that
the regression was significant at the . 975 level with 10 and 24 degrees
of freedom.
That is, the hypothesis that all the
values equal zero was rejected.
(3
coefficient
The observed F value for the com-
plete regression was 3. 2093 and the critical value was 2 64.
Elasticities were calculated at the means and the results should
only be interpreted as meaningful for small incremental changes about
those mean values.
3
Variables accounting for the influence of hypothesized substitute
goods in this model include the price of medium salmon to retailers,
the price of "large" trout to wholesalers, and the price of pansize
salmon to retailers.
Of these potential substitutes only pansize
3
"Price elasticity of demand or elasticity of demand is the
proportional change in the consumption of a good divided by the proportional change in the price of the good" (Ferguson and Gould, 1975,
p. 46).
Q
P
^
ap Q
"Cross-price elasticity of demand measures the relative responsiveness of quantity demanded of a given commodity to changes in the
price of a related commodity" (Ibid- , p. 103).
ap
5 /
Y
Q
P
x
y
8Q
QQ
=
xy
=
5
aP
y
P
_X
Q
x
"Income elasticity of demand is the relative responsiveness of
quantity demanded to changes in income" (Ibid. , p. 104).
ao ,av -_ „n income
.
_ ao Y
- — - .
T
- /
45
salmon failed to enter the equation at a statistically significant level,
as indicated by calculated t-values.
Medium salmon and "large"
trout appear to be strong substitutes for rainbow trout.
The cross-
price elasticities associated with rainbow trout and each of these
substitute commodities indicate an "elastic" relationship.
case of medium salmon the variable
P
rms
w
4
In the
represents what the
wholesaler "receives" for selling medium salmon.
Thus the elastic
cross-price relationship implies that, for example, a 1% increase in
the price received by the wholesaler for medium salmon will result
in a greater than 1% decrease in the quantity of rainbow trout he will
demand from the broker, ceteris paribus.
The variable
P
wirgt
represents what the wholesaler must "pay" for the commodity.
The
positive elastic relationship, in this case, indicates that, for example,
a 1% increase in the price the wholesaler faces for "large" trout will
result in a greater than 1% increase in the quantity of "rainbow" trout
he demands, ceter is paribus.
The variables representing own-price did not enter the equation
at statistically significant levels.
4
However, the signs on the
If the cross-price elasticity of demand > 1 the relationship is
said to be "elastic". That is, for example, a 1% change in the price
of commodity Y will result in a greater than 1% change in the quantity of X demanded, ceteris paribus- The relative direction of these
changes depends upon the specific relationships which exist between
the two commodities, i. e. , substitutional, complementary or
independent-
46
coefficients and magnitudes of the elasticities are consistent with
expectations.
Personal disposable income appears as a significant
explanatory variable in the analysis of wholesale demand.
The
resulting estimated elasticity indicates that the relationship between
quantity of rainbow trout demanded by wholesalers and personal disposable income in the Los Angeles SCSA is positive.
Indeed, the
deinand for rainbow trout appears to be income elastic.
That is, as
personal disposable income for this area increases by, for example,
1. 0% the quantity of rainbow trout demanded by wholesalers will
increase by 2. 47% ceter is par ibus ■
Certain writers have suggested that commodities can be
classified as "necessities" and "luxuries" on the basis
of income elasticity. If income elasticity is very low
(certainly less than one), quantity demanded is not very
responsive to changes in income. Consumption remains
about the same irrespective of income level. This suggests that the commodity in question is a "necessity". On
the other hand, an income elasticity greater than one indicates that the commodity is more or less a "luxury"
(Ferguson and Gould, 1975).
According to Ernst Engels, the income elasticity of demand for food
as a general category is very low, but specific items might very well
be perceived as "luxury" goods.
This result tends to lend credence
to information received from industry sources concerning consumption patterns for rainbow trout.
The general impression has been
that Idaho rainbow trout is a premium food product primarily
47
consumed as a luxury item within the restaurant and institutional
distribution network.
Interestingly, this consuming market is
exactly identical to the "targeted market" identified in the promotional
strategy of the pansize salmon producers.
Further, the descriptive
term "luxury food item" appears in some of the pansize salmon
literature.
48
VI.
FURTHER ANALYSIS AND HYPOTHESIS GENERATION
Despite solid assurances of cooperation from key industrysources, the previously mentioned data, upon which further analysis
had been anticipated, have not been forthcoming.
As a result of this
unfulfilled commitment by one or two of the most significant trout
producers the remainder of this analysis is limited to hypothesis
generation in hopes that, perhaps, future studies might succeed in
obtaining critical data and proceed to quantitatively evaluate the
merits of the several hypotheses presented below.
As was explained in the preceding chapter, it had been hoped
that a demonstrative presentation of a simple model might serve as a
tool with which to persuade others to contribute data.
Thus the single
equation, quantity dependent model was estimated using ordinary
least squares techniques.
However, it had been strongly felt that the
more correct specification of the Los Angeles market would require a
system of equations to account for the simultaneous determination of
price and quantity in the market.
The assumption implicit in the
single equation model is that price is somehow determined outside the
model.
For example, one explanation might be that Idaho producers
set a price and then allow the market to define an equilibrium quantity at that price.
Another explanation is that the price in the L.A.
market in preceding periods determines the quantity available for
49
distribution through the marketing network in the time period being
observed.
Subsection A: The Money Illusion, a Simultaneous
Equations Model
None of these alternative explanations seemed very satisfactory
based upon the results of extensive investigation and interviews with
industry sources.
In fact, every indication was that rainbow trout
characteristically has been a highly price-competitive commodity in
the Los Angeles market.
This observation led to the specification of
a simultaneous equation system in which it was implicitly hypothesized
that the price of rainbow trout and equilibrium quantity are interdependent and, thus, simultaneously determined in the market place.
That is, both own-price and quantity are endogenous variables.
As
has been discussed earlier, the remainder of the price variables in
this specification are assumed to be predetermined-
5
This follows
from the assumption that the quantity of rainbow trout demanded by
5
Foote (1958) defines an endogenous variable "as one that is
correlated with the unexplained residuals in the structural equation in
which it appears, " while a predetermined variable "is independent of
the unexplained residual in the structural equation in which it
appears". Kmenta (1971, p. 532) notes that "predetermined variables
can be subdivided into exogenous and lagged endogenous variables.
The values of the exogenous variables are completely determined
outside the system under consideration, whereas the values of the
lagged endogenous variables are represented by the past values of the
endogenous variables of the model".
50
wholesalers in time
"t"
depends, not only upon the current price
of rainbow trout and its hypothesized substitutes at the brokerage
level, but also upon the most recent available information concerning
wholesale to reatial prices.
This implies a lag, in this case one time
period, in the several price series accounting for substitutes and
own-price behavior in the wholesale to retail sector.
The system of structural equations for this particular
specification include:
^Wt -
dt
X
0
-0^ _L
1 bt
+ Q
T}1"1113
2 wt_1
6Yt+a7Dtmt
t
t
X
+ a
8D2
TDW-Crgt X
■Drt
3 bt
T-,rPSS X
4 wt_1
5 wt_1
+ a
9D3+Q10D4
t-1
0t
QWt
=
QT
=
Ql
*\
s
d
trout
in aggregate, or alternatively, in per capita terms the system
becomes:
Qwt
d
_
,
_wt
T-.rrns
n^pss
^rt
,
Tr /lvT
—— = a0 + a.P,
+ a.P
+ a _P c
+ a .P
+ a.Y/N^
N
0
lb
Zw^.
3w.
4 w .
5
t
t
t
t-1
t-1
t-1
^imt ,
^ ,
+Q D
6Dt
7 2 +
+ Q
^
8D3
a
^ ,
9D4 +
+ Q
T-,wirgt
10Pbt
Q
51
Qwt
wt
s
Nt
Qwt
d
Nt
total
trout
Nt
where the variables in the demand equations are specified as before
(see V, page 40 and 41), and
in time
N = population of Los Angeles SCSA
t.
_,
,
.
.
.
, ,. .
The supply equation contains, in addition to
T-,wt
P
:
t
Q
wt
S
or Q
wt
/N = quantity of rainbow trout supplied to brokers
St
for sale in L. A. in time t, in total or per capita terms
wnyt
P,
= price of rainbow trout from broker to wholesaler in time
b
t
t for the Fulton Fish Market, New York (deflated)
N = population of Los Angeles SCSA in time t
T = variable "time" accounting for supply shifters not otherwise observable given the available data.
All prices were deflated by the W.P.I, for "Farm products, processed foods and feeds, " U.S. Bureau of Labor Statistics.
Two runs
were made using, in the first, a non-deflated income variable,
and
By specifying the model in terms of a non-deflated personal
disposable income variables, either per capita or total, the implicit
argument might be framed in terms of a behavioral model. It can be
52
in the second an income variable deflated by the consumer price
index (CPI), U.S. Bureau of Economic Analysis.
rr,,
• •. ,-,
The identity
^wt
^wt
^total
Q,
= Q
= Q
d
s
trout
...
.,., .
specifies
an equilibrium
condition in which quantities supplied are exactly equal to quantities
demanded.
As noted earlier, all available information indicates that
no significant stocks of rainbow trout are held at any of the three
marketing levels in the L.A. distribution network.
Stocks do appear
to exist at production/processing sites from which shipment orders
originating in L.A. can be met within 24 hours of receipt, according
to industry sources.
In the supply equation the inclusion of the New York Fulton Fish
Market price series for rainbow trout was intended to account for Los
Angeles supply shifts resulting from price fluctuations in alternative
market areas.
The variable
T
represents time and as explained
above was intended to account for supply shifting forces at work in the
market which are not otherwise observable given the data limitations
discussed earlier.
The rationale for the inclusion of the demand
equation variables has been described in the preceding section and
readers are referred there for further clarification.
hypothesized that consumers determine their consumptive patternsj
in the short run, principally upon their perceived or "nominal"
income. That is, "money illustion" tends to dominate short term
consumptive behavior.
53
While it would have been possible at this point to estimate
coefficients for both the demand and supply equations using one of
several econometric estimation techniques, for the purpose of this
analysis, only the demand equation was actually quantitatively evaluated.
The results of both the deflated and non-deflated income speci-
fications can be seen in Appendix B and C respecitvely.
A two stage
least squares technique was employed in order to account for the
simultaneity of own-price and quantity in the hypothesized model.
This method was selected over alternative techniques on the basis of
several criteria, chief among them being the suitability of 2SLS as
an estimation method for an over identified equation.
7
In addition,
7
An equation is said to be over-identified, "when the number of
the unrestricted coefficients exceeds the number of the restricted
parameters and there is no unique solution" (Kmenta, 1971, p. 540).
The criteria for determining identification of an equation includes rank
and order conditions. The order condition requires that the total
number of exogenous variables (K*) excluded from the equation of
interest must be at least as great as the total number of endogenous
variables (GA) in the given equation minus one. Or,
K* >GA - 1
If K* = G - 1 the equation is exactly identified
If K* > G - 1 the equation is over-identified
If K* < GA - 1 the equation is under-identified.
The rank condition is somewhat more complex. However in essence,
the rank condition requires that there be at least one independent
equation for each endogenous variable in the system.
In this case the rank condition is satisfied, however, the
demand equation is over-identified on the basis of the order condition.
54
the ready availability of the two stage least squares technique package
on the available electronic data processing system increases the cost
of using one of the alternative, more mathematically cumbersome
methods of estimation.
In addition none of these alternative methods
demonstrate a clear advantage over two stage least squares, according to Kmenta (1971).
A third case can be made for employing two stage least squares
in as much as it "does not require complete knowledge of the whole
system; all that is required is a listing of all predetermined variables
and their sample values" (Kmenta, 1971, p. 562).
The results of the simultaneous models were somewhat disappointing.
In the first case, non-deflated
Y,
price of rainbow trout
from broker to wholesaler was regressed against all the predetermined variables in the model in the first stage.
A
a predicted own-price,
P
This process yielded
wt
, broker to wholesaler, which was then
t
wt
included as an explanatory variable, in place of the observed P, ,
t
in the second stage estimation of the demand equation with per capita
quantity of rainbow trout as the dependent variable.
8
g
Subsequent to the above cited per capita specification the model
was re-evaluated with the substitution of "total" for "per capita"
demand and income figures. The results of this run were similar to
those reported above. Where the two specifications do vary slightly,
the per capita appeared to be the superior delineation by R2 criteria.
55
The first stage regression was highly significant with the
25. 8910 F-statistic for the total regression exceeding the critical F
value by a magnitude of more than eight, at the . 05 level, with 11,
25 degrees of freedom.
The
R
2
for the first stage was .9139.
The second stage regression in which per capita quantity was
the dependent variable, yielded somewhat less encouraging results.
The F-statistic for the complete regression, 3. 3681; with 10, 25 d.f.,
exceeded the critical F value 2. 24;
a = 0. 05.
Furthermore, the
signs on several of the coefficients in the second stage regression
were not consistent with the previously hypothesized relationships.
An example is the relationship between rainbow trout demanded and
the price of pansize salmon at the wholesale to retail level.
Pansize
salmon has been hypothesized, from the inception of this study, to be
a strong potential substitute for rainbow trout in the market place.
Pansize salmon does enter the second stage equation at a statistically
significant level based upon the t-statistic-
However, the sign on the
coefficient of pansize salmon is positive, thus implying that as the
price received by the wholesaler for supplying pansize salmon to his
retail customers rises, the quantity of rainbow trout he, the wholesaler, will demand will increase as well.
with the expected relationship.
This is clearly inconsistent
More disturbing than this inconsist-
ency is the "incorrect" sign on the coefficient predicted for own-price
from broker to wholesaler.
This variable enters the equation at a
56
significant level, based again upon the associated t-statistic.
However, contrary to the inverse or negative price/quantity relationship expected, a positive relationship emerged.
trout from the wholesaler to retailer in time
model at a significant level.
The price of rainbow
t-1
failed to enter the
Dropping this variable on grounds that
wt
rt
P,
and P
are highly intercorrelated, thereby making it difb
w, ,
t
t-1
ficult to sort out the separate effects of each, failed to change the
estimated sign of the coefficient on
P
wt
.
The Durbin-Watson statis-
t
tic, used as a mechanism for assessing auto-regress ive tendencies in
the data, is in the indeterminant range in this case, 1. 6932.
The personal disposable income variable, non-deflated,
entered the regression at a highly significant level and with the
expected positive sign on the coefficient.
However, care must be
taken in interpreting the significance of this result based upon the
accompanying coefficients.
The remaining exogenous or explanatory
variables included in this specification failed to enter the equation at
significant levels.
However, their exclusion from the model in sub-
sequent runs only tended to aggrevate the problems and substantially
reduce the regressions' significance.
57
Subsection B: A Simultaneous Equations Model Expressed in
"Real" Price and Income Parameters
The alternative specification of this simultaneous, quantitydependent model included only a single modification.
This involved,
as noted above, the use of a deflated personal disposable income
figure,
or
Y*
Y/N ,
or
Y*/N ,
rather than the nominal income figure,
used previously.
Y
The model was specified, as was the
previous one, in both "total" and "per capita" terms, for purposes of
comparative analysis.
The results of the "per capita" run were only
slightly different from the "total" equation results and thus only the
per capita results are addressed directly in this subsection.
Again,
complete results of this specification are included in Appendix C.
Once more the two stage least squares technique was employed.
In the first stage own-price (broker to wholesaler) was regressed
against all of the predetermined variables in the system of equations,
this time including a deflated per capita disposable income variable,
in place of the nominal income variable.
The resulting predicted
^wt
P , was introduced into the second stage quantity dependent
t
wt
equation, in place of the observed P, , as an explanatory variable.
t
The remainder of the predetermined variables are identical to those
price,
contained in the preceding model except for the substitution of the
deflated per capital disposable income variable, identified above.
The results were very interesting.
58
In the first stage estimation the regressions significance,
based upon the F-statistic, and the "goodness of fit" or explanatory
capacity, based upon the
R
2
statistic, were ver y slightly reduced
from the same figures in the preceding model.
The
R
2
declined
from . 9193 to . 9098 and the F-statistic for the total regression, first
stage, dropped from 25. 891; (11, 25d.f. ) to 22. 9266; (11, 25d.f. ).
However, it was the second stage results which caused the greatest
interest.
While the second stage F-statistic and
R
2
did improve
marginally, and the Durbin-Watson statistic remained virtually
unchanged, there was a substantial difference in the individual significance of several variables based upon their t-values and the signs
of their respective coefficients, when compared to the earlier model.
Specifically, ever y var iable except that accounting for fourth quarter
seasonality shifts in demand, entered the second stage regression at
significant levels.
Under this formulation, the price of large trout at
the brokerage level appears to be positively correlated with the
dependent variable. It also would seem to be a highly significant
explanatory variable with regard to the variation observed in quantity
of rainbow trout demanded.
That is, for example, as the price of
large trout falls at the brokerage level, presumably due to an
increase in supply, wholesalers tend to substitute away from rainbow
trout, thus reducing the quantity of rainbow trout demanded, ceteris
par ibus.
This is precisely the relationship hypothesized and from a
59
comparative standpoint, just the opposite of the results of the
preceding model.
Recall that
P
wirst
in the first model failed to
t
enter the regression at a statistically significant level. Similarly, the
b
price received by wholesalers for pansize salmon entered the second
regression at a highly significant level and, unlike the results of the
first model,
P
I" D S S
did so with the expected negative sign.
Thus
W
t-1
the implicit relationship, in this second case, between the price of
pansize salmon, wholesale to retail, and the quantity of rainbow trout
demanded by the wholesale sector, is one of substitute input factors
into the wholesalers' production function.
For example, as the price
of pansize salmon the wholesaler can command from the retail level
decreases, he will tend to substitute away from pansize salmon and
increase his demand for rainbow trout, ceter is par ibus.
The presence of Japanese trout in the Los Angeles market does
appear to have some negative influence upon the quantity of rainbow
trout demanded, based upon the results of this second specification.
Once again, this is contrary to the results of the regression of the
first model, but consistent with a prior expectation inasmuch as the
influence, while statistically significant, appears to be rather
negligible in absolute terms.
This might be interpreted as accounting
for some of the fluctuation in the small quantities of domestic rainbow
trout which pass through the distribution network to the "market
outlet" level where they interact with the less expensive Japanese
60
product.
If the imported fish were readily available, and at the
character istically lower per unit price, then it might be expected that
"market outlet" purveyors would substitute away from the more costly
domestic product.
When supplies of Japanese trout tighten, the void
is filled by increased demand for domestic product.
This would
surely have some observable effect on levels of demand for rainbow
trout within the L.A. distribution network.
As the preceding chapters
have attempted to establish, the magnitude of the effect would be
expected to be relatively insignificant.
In a sense, the demand rela-
tionship specified above may be seen as the sum of two demands,
namely "market outlet" and "food service", with the former involving
relatively small quantities of Idaho trout-
The Japanese import vari-
able may account for shifts in the "market outlet" demand.
The prices of rainbow trout, both
from broker to wholesaler and
P
rt
& wt
P
(first stage predicted),
b
t
from wholesaler to retailer,
W
t-1
enter the regression at statistically significant levels.
in contrast to the earlier specification, they do
sized signs on their coefficients.
BO
Furthermore,
with the hypothe-
The price of rainbow trout received
by the wholesaler is positively correlated with the quantity of rainbow
trout demanded at the brokerage level.
As the price the wholesaler
is able to command for the product increases, for example, the
quantity he demands from the broker will similarly increase, ceter is
paribus ■
61
The predicted price of rainbow trout at the brokerage level is a
very significant explanatory variable, both in terms of the t-value and
the magnitude of the variables' coefficient, in this quantity-dependent
second stage.
The relationship, as hypothesized, is negative, imply-
ing that, as the price of rainbow trout at the broker to wholesaler
level increases (decreases) the quantity of rainbow trout demanded
will respond in the opposite direction.
The binary variables intended to account for seasonal shifts in
demand do, in two of the three cases, enter the regression at significant
t
levels.
The second quarter of the calendar year appears to
reflect a period during which consumption of rainbow trout is on the
decline.
This seems to be in agreement with industry perceptions.
The first quarter characteristically is "the" big seafood season as
the Lenten period is observed.
Following the end of the religious
observance there tends to be some adjustment away from fish, in
favor of alternative food protein items.
Thus a more complete
specification might have included consideration of these items
explicitly.
The decline in demand for rainbow trout in this period
probably reflects a more general downward trend in seafood consumption.
The third quarter binary variable also enters the equation at a
significant level and would appear to reflect a moderate upturn in
demand following the second quarter decline.
The fourth quarter
failed to enter the equation at a statistically significant level.
62
Unexpectedly the two remaining variables, namely the price of
medium salmon, wholesaler to retailer, and deflated per capita
personal disposable income, while entering the regression at statistically significant levels, do so with "erroneous" signs.
That is, in the
case of the price of medium salmon, an hypothesized substitute for
rainbow trout, the coefficient is significantly different from zero and
positive.
This is contrary to the substitutional hypothesis.
Per capita personal disposable income, deflated, enters the
regression with a negative signtions for this result.
There are several possible explana-
The negative coefficient may be the result of
statistical problems within the data.
It is, of course, not theoreti-
cally inconceivable that, over some range, rainbow trout is, in fact,
an inferior good-
Ignoring the layman's negative connotation of the
term "inferior good", the economic implications are, simply stated,
that as income increases the satiation point for the consumption of
rainbow trout decreases.
This is not so difficult to imagine when one
recalls that income can be perceived as a proxy for "all other goods".
In this sense for example, as a consumer increases consumption,
including that of alternative food proteins such as beef, pork, poultry
and other seafoods, the utility he obtains from consuming rainbow
trout may be diminished over some range.
Alternatively perhaps,
over the long run, recalling the very short time series available for
use in this analysis, one might hypothesize the true relationship is
63
actually positive.
Thus if sufficient data are available one would, in
fact observe a positive income/quantity relationship.
In any event, the presence of the "unexpected" coefficients on
these two variables casts a shadow on the reliability of the remaining
signs and coefficients.
Nonetheless, in light of the serious, and at
the present unresolvable, data problems the results of this specification are quite encouraging.
Any final resolution of the model's
validity must await the broader participation and cooperative support,
in terms of data provision, of the Idaho food fish industry.
Perhaps
the responsibility for the absence of broader industry collaboration
can be traces to a failure to make clear the project's goals and objectives.
Furthermore, as has been mentioned previously, while
oligopolistic in structure the history of intense price rivalry of this
industry may have contributed to the reluctance by making the perceived costs very high relative to the potential benefits.
Subsection C; A Single Equation Price
Dependent Alternative Specification
The preceding model specifications are consistent with the vast
majority of available information obtained throughout the interview,
data acquisition segment of this study concerning the L. A. marketFurther, the preceding models are consistent with the theoretical
framework hypothesized within this analysis.
They are also based on
64
the best estimates of the Los Angeles market structure in terms of
the author's perceptions of that distribution network.
However, in
response to a somewhat divergent view of the L.A. market, on the
part of one of the Idaho producers interviewed, and in an effort to
examine the validity of all available leads, a third model was specified
and quantitatively evaluated.
In this case the price of rainbow trout
at the brokerage level was identified as the dependent variable in a
single equation model.
The quantity of rainbow trout entering L.A.
was assumed to be a predetermined variable under this specification.
The basis for this assumption was drawn from an interview, mentioned
above, during which the L.A. market was described as a traditional
dumping ground for excess supplies of rainbow trout, from domestic
farms.
In essence the prices in alternative U.S. markets determine
the quantities which will ultimately find their way into the Los Angeles
distribution network under this specification-
Once the producers
have determined the magnitude of their shipments to alternative markets the remainder of the available supply is either used to replenish
stocks or is shipped into L.A. where it is sold for whatever the product will bring.
This argument seemed somewhat inconsistent with the history of
highly price competitive market behavior in L.A. , cited earlier in
this study.
However, at this stage of the analysis the primary goal
was one of hypothesis-generation.
Completeness required that this
65
specification be explored.
The exact specification of this equation is
as follows:
^wt _
b
d
0
IN
Y
2N
t
t
+ acPrt
5w
-D^pss ,
Sw,
t-1
-crms
4w.
t-1
+ a.D, + a_D, + a QD .
62
73
84
where the variables are as specified in the preceding subsection.
A complete presentation of the results of the ordinary least
squared (OLS) regression appear in Appendix D.
the model does yield some interesting results.
Briefly, however,
For example, the
price that wholesalers received in the preceding period,
t-1,
is a
highly significant explanatory variable in terms of accounting for the
variation associated with the brokers price of rainbow trout in time t,
based upon the t-statistic-
Further, this relationship is positive, as
one would expect to observe based upon the hypothesized interaction of
the two price variables.
The price of pansize salmon from whole-
saler to retailer also enters the regression at a highly significant
level, again based upon the t-statistic at a . 90 value.
The negative
coefficient on pansize salmon's price is also consistent with the
expected substitutional relationship.
That is, as the price of pansize
salmon received by the wholesaler increases, for example, the
wholesaler should substitute away from rainbow trout, in favor of the
66
more profitable pansize salmon, ceter is paribus, thus necessitating
a reduction in the price of rainbow trout by the broker in order to
move the available supply.
Per capita quantity demanded of rainbow trout, while entering
the regression at a significant level, does so with a positive coefficient.
This is contrary to the expected or hypothesized relationship
between the predetermined quantity variable and the broker to wholesale price of rainbow trout.
The positive sign on the coefficient
would imply that as supplies of rainbow trout increased (decreased),
into the Los Angeles market, the price of rainbow trout would likewise increase (decrease).
This is inconsistent with both the predic-
tions of the economic theoretical constructs and the generally assertained perceptions of the workings of the L. A. market held by those of
its participants surveyed.
Similarly, the relationship between per capita personal disposable income, and the dependent variable implied by the negative
sign on the former's coefficient is not consistent with expectations.
Seasonality appears not to be a significant explanatory factor in this
specification based upon the OLS results.
This is particularly inter-
esting in light of the very strong opinion to the contrary held by the
industry member who's statements prompted this particular model
configuration.
Despite the somewhat improved
2
R ,
. 77, and the
consistent coefficients on most of the variables in this price dependent
67
model, it remains the opinion of the author that the quantity dependent,
simultaneous price /quantity model, specified in the preceding subsection, most nearly approximates the actual Los Angeles distribution network's behavior.
However, this can be explored further when and if subsequent
research is done in this field, and assuming industry members provide adequate support in terms of cooperation and data accessibility.
68
VII.
SUMMARY AND CONCLUSIONS
Because not all of the major participants in the Los Angeles
market were willing or able to provide needed data, it has not been
possible to provide a thorough description of that distribution network.
Despite these data limitations however, an initial effort has been
made to quantitatively approximate and statistically evaluate the market for rainbow trout and the role of its hypothesized substitutes,
including among them, pansize salmon.
In light of the results of the
analysis discussed in Chapter VI some interesting and potentially
significant observations can be made.
Several models were examined.
The results of testing the hypotheses of two of these models, namely
the alternative simultaneous equations specifications, are particularly
interesting.
In the first model, two stage least squares techniques
were used to estimate the parameters of a quantity-dependent demand
equation.
This equation included, predicted own-price (broker to
wholesaler) deflated, price of trout (wholesale to retail) lagged one
period and deflated, the prices of several substitutes also lagged and
deflated, personal disposable income, non-deflated, and several
binary variables accounting for seasonality and the presence of
Japanese trout in the market place.
(All prices were deflated by the
Wholesale Price Index for Farm Products, Processed Foods and
Feeds.)
69
The second specification is identical to the first with the single
exception that the personal disposable income variable is deflated by
the Consumer Price Index.
As the reader may observe, the substan-
tial improvement in the second simultaneous equations specification
over the first, as measured by the
2
R ,
F and t
statistics,
resulting, as just mentioned, exclusively from the deflation of the
income variable, gives rise to several interesting questions.
If, for
example, one had chosen to specify the first simultaneous model
totally in terms of "nominal" figures, instead of incorporating deflated
prices and non-deflated income variables, would the results have been
more closely akin to those obtained in the second model? It seems
reasonable, since under the assumption of the first model, the consumer behaves under a "money illusion" concerning his relative
economic well being.
When the analyst looks at monthly purchases,
as in this analysis, the significance of "real" income and "real"
prices may be diluted because the consumer is not able to fully perceive and internalize these "real" changes in his relative economic
status.
Perhaps then, in a future specification one might attempt to
incorporate a lag of one or more time periods in order to attempt to
account for this adjustment to "real" parameters on the part of the
consumer.
Alternatively, perhaps by aggregating the data in such a
way as to minimize the influence on consumptive behavior of this
adjustment period, one might be more successful in isolating the
70
actual demand relationships at work in the market place.
The use of
longer observation periods, quarterly or semi-annual increments,
might also reduce the impact of non-quantifiable influences which may
distort short run behavior, but which have no long run influence on
consumer demand.
Evaluation of these modifications of the basic
simultaneous equations model, presented above, must await additional
inputs in terms of historical time series data on the product.
Based upon the results of the analysis several conclusions might
be inferred.
To begin with, there seems to be a very strong substi-
tutional relationship implied, by the results of the simultaneous equations model, between rainbow trout and pansize salmon.
Data
limitations prevent a quantitative estimation of the magnitude of this
relationship.
However, the existence of such an interaction between
the two aquaculturally produced food fish perceived by industry members in the earliest stages of this analysis, seems to be confirmed.
Further, seasonality appears to be an influencing factor in the demand
for rainbow trout.
In terms of the second model, Japanese trout appear to have
only a marginal influence on the demand for rainbow trout.
However,
as the imported product becomes more scarce in the U.S. market,
domestically produced trout will probably come under increasing
demand to fill the void.
This may be expected to amplify the compe-
titive interaction between "food service" and "market outlet"
71
consumers for the domestic product.
Beyond this, the analysis of this market relationship must
await more complete data series.
It is the fervent hope of the author
that this study serves primarily to identify critical issues, ask the
relevant questions, and pose provocative alternative explanations as
a means of informing all interested parties of the potential benefits
forthcoming if such questions can be objectively answered.
In line
with this, it seems indisputably clear that further research into the
market behavior and potential of aquaculturally produced food proteins, and particularly that of rainbow trout and pans ize salmon which
have generated so much recent interest, cannot be too soon in
coming.
72
BIBLIOGRAPHY
Batie, Sandra S. 1974. The United States' importation of fishery
products: an econometric case study of the southern Atlantic
and gulf shrimp industry. Ph.D. thesis. Oregon State
University.
Bardach, J. E. , H.H. Ryther, andW.O. McLarney. 1972.
Aquaculture: the farming and husbandry of freshwater and
marine organisms. Wiley-Inter science. New York.
Bell, Fredrick W. and Jared E. Hazelton (eds. ). 1967. Recent
developments and research in fisheries economics. Oceana
Publications, Inc. New YorkBoulding, Kenneth E. 1966. Economic Analysis, Volume I, Microeconomics. Harper and Row, Publishers. New York.
Brandow, G.E. 1961 (August). Interrelations among demand for
farm products and implications for control of market supply.
The Penn. State University, College of Agriculture, Agricultural Experiment Station, University Park, Penn. Interregional
publication for the State Ag. Exp. Station, Bulletin 680.
Brannon, Ernest, Terry Nosho, Ray Nakatani, and Ernest Salo (eds.).
1973. Workshop on salmonid aquaculture - a summary report.
A Washington Sea Grant Program report. Division of Marine
Resources. University of Washington.
Brown, William G. , Doug Larson, Richard S. Johnston, and Roy J.
Wahle. 1976 (February 13). Improved economic evaluation of
commercial and sport-caught salmon and steelhead of the
Columbia river. Oregon State University.
California Statistical Abstract.
Finance.
1975.
California Department of
Charoenkul, Vilailuck. 1971. Analysis of demand for canned pink
salmon. Master's thesis. Corvallis, Oregon State University.
Crutchfield, James A. and Arnold Zellner. 1963. Economic aspects
of the Pacific halibut fishery. Fisheries Industry Research,
Volume 1 - Number 1. United States Government Printing
Office, Washington.
73
Crutchfield, James A. and Giulio Pontecorvo. 1969- The Pacific
salmon fisheries - a study of irrational conservation. The
Johns Hopkins Press. Baltimore, Maryland.
Current Fisheries Statistics No. 6700. Fisheries of the United
States, 1974 March, 1975, U.S. Department of Commerce
National Oceanic and Atmospheric Administration. National
Marine Fisheries Service.
Deloria, Vine, Jr. 1973 (Nov. -Dec. )• The tribe that was made to
be fishermen. American Fishes and U.S. Trout News. pp. 13
20.
Ferguson, C. E. and J. P. Gould.
IRVIN. Homewood, 111.
1975.
Microeconomic Theory.
Foote, R. J. 1958. Analytical tools for studying demand and price
structure. U.S. Dept. of Agriculture. Agriculture handbook
No. 146. Washington, D-C 217 p.
Hanson, Joe A. (ed. ). 1974. Open sea mar iculture.
Hutchinson and Ross. Strandsburg, Penn.
Dowden,
Jones, Walter G. 1967 (Jul. -Aug. ). Here's the market potential for
farm raised fish. American Fishes and U.S. Trout News.
Kane, Thomas E. 1970. Aquaculture and the law. Sea Grant Technical Bulletin #2. Miamis University of Miami Sea Grant
Program.
Klontz, George W. and John G. King. 1975. Aquaculture in Idaho
and nationwide. Idaho Department of Water Resources.
Research Technical Completion Report.
Kmenta, Jan. 197 1. Elements of Econometrics.
lishing Co. New York.
Macmillan Pub-
Lannan, James E- 1975. Netarts Bay chum salmon hatchery - an
experiment in ocean ranching. Oregon State University Sea
Grant College Program. Agricultural Experiment Station.
Bulletin no. 621.
Neter, John and William Wasserman. 19 74.
cal models. IRVIN. Homewood, 111.
Applied linear statisti-
74
Richards, Jack A. , Conrad V. W. Mahnken, and George K. Tanonaka,
(invest.). 1972. Evaluation of the commercial feasibility of
salmon aquaculture in the Puget Sound. Northwest Fisheries
Center. National Marine Fisheries Serivce. National Oceanic
and Atmospheric Administration. Seattle, Washington.
Robinson, Clay M. 1972. A marketing plan would stabilize trout.
American Fishes and U.S. Trout News, l6(6):19-20.
Robinson, Clay M. 1972. A holiday with trout on the menu.
American Fishes and U.S. Trout News, Vol. 17, pp. 12-13.
U.S. Bureau of Labor Statistics. Economic Indicators, p. 22.
U.S. Government Printing Office. Washington, D. C. 1975.
Wagner, Louis C. 1972. An evaluation of the market for pan-sized
salmon. Prepared for National Marine Fisheries Service,
NOAA, U.S. Department of Commerce. 50 p.
Waugh, Fredrick V. and Virgil J. Norton. 1969. Some analysis of
fish prices. U.S. Bureau of Commercial Fisheries. Working
paper #22. Washington, D.C. 194 p.
Wonnacott, Ronald J. and Thomas H. Wonnacott.
metrics. John Wiley & Sons. New York.
1970.
Econo-
Wood, William R. 1970 (June). A demand analysis of processed
salmon from the west coast. Master's thesis. Corvallis,
Oregon State University. No. leaves 78.
APPENDICES
75
APPENDIX A
For the convenience of the reader the economic model appearing
on pages 40 and 41 is reproduced below as are the definitions of the
model's components.
The economic model specified was as follows;
! p prt prms
ptm D
D Y Dimt
of
r. ' > <'rgt>prpss
* ?Z34
' D,. ^ '
dbwwb
w '
and assumes the identity;
Wt
t0tal
Qf
d = Qs = Q
recalling the colon is read "depends on", a comma is read "and",
Q
wt
= quantity of rainbow trout supplied by producers to the
L. A. market
wt
Q, = quantity of rainbow trout demanded by L. A. wholesalers
at brokerage level
wt
P, = price of rainbow trout from broker to wholesaler
b
rt
P
= price of rainbow trout from wholesaler to retailer
w
P
= price of rainbow trout from wholesaler to market outlets
r ms
= price of medium salmon from wholesaler to retailer
w
w^rgt
P,
= price of large" trout from broker to wholesaler
P
P
r TDS s
= price of pansize salmon from wholesaler to retailer
w
76
Y = personal disposable income in the L.A. (SCSA); not
deflated
D
= binary variable indicating presence or absence of imported
trout in market
D7,D,,D
£
J
rr
= binary variables included to account for seasonality
shifters on quarterly basis.
The econometric equation actually estimated using Ordinary
Least Squares is as follows;
QT1 = o + a Prt
+ a PrmS + a D. + a D, + a D + a Y
d
0
lwn
Zwn
32
43
54
6t
t
t-1
t-1
+ Q D
imt ,
7 t
^wt ,
+ a P
8 bt
+
^wirgt ,
^bj
+ a
^rpss ,
10PJt_1
+ a
^mt ,
llPt-l
+
^
where variables are defined as indicated in the economic model, the
present month is
t,
the preceding month
t-1
and
[>■
is a
stochastic error term.
The results of the analysis are:
Qj = -24414 + 7675. 2P
- 12907PrmS + 2395. 5D
t
(.258) Wt-1
(-1.43)Wt-l
(1.22)
..5912
-1.1
+ 5039. 3D
(2.65)
+ 3935. OD
(1.68)
+ . 00060099Y + 17. 57lDimt - 2374200P^t
t
(3.65)
(.043) t
(-.797)
t
2.47
-1.17
+
77
+ 3054400P^
(1.67)
t
1.22
gt
+ 4282ZPrpSS + 758580Pm*
(. 143) Wt-1
(.392) t"
.00488
R2 = .562
where the t-values are in parenthesis below the coefficients, and
single-tailed tests were used.
below the
t
statistics.
The elasticities are immediately
The t-values are statistically significant at
the 90% level with 25 degrees of freedom if they exceed the critical
value 1.316.
The 95% significance level critical
t
is 1.708.
The period of analysis is May 1972-December 1975, inclusive.
The number of complete observations with which the analysis was
carried out totals 36.
The least squares assumptions made with regards to
p.,
in
this specification, as well as the error terms in the subsequent
alternative formulations include:
1. E(fi.) = 0
2. E(HL2) = cr2
3. E^.n.) = 0;
(i 4 j)
4. Each explanatory variable is nonstochastic with values fixed
in repeated samples.
5. The number of observations exceeds the number of coefficients to be estimated.
78
6. No exact linear relation exists between any of the
explanatory variables.
For purposes of testing one additional assumption is required.
7.
IJL.
i
is normally distributed
where the subscript
term.
i
indicates the
ith
observation of the error
79
APPENDIX B
The first simultaneous equation's specification included a nondeflated personal disposable income variable as explained in Chapter
VI, Subsection A.
QWt
d
-^
.
%
The specification of this system was as follows:
_wt .
-nrins ,
-n^pss ,
^rt
.
l bt + "2^./ "S^j + "^w^j +
+ a P
,._. ,.,
5Y/Nt
a
%Dr+-7^+asDs+v*+x+aiopr'rgt
+
^wt
Q
s
N
t
^o^r^r'^s^*
^wt
^wt
^total
s
_ d _ trout
N
N
~
N
t
t
t
where the variables in the demand equations are specified as before,
(see V, page 40 and 41 or Appendix A, page 75),
tion of Los Angeles SCSA in time
t,
and
K
and
and
N
6
= popula-
are stochas-
tic error terms.
The supply equation contains, in addition to
P
wt
".
t
Q
wt
S
t
or
Q
wt
/N
ST.
= quantity of rainbow trout supplied to brokers
for sale in L. A. in time t,
in total or per capita terms
80
wnyt
P,
= price of rainbow trout from broker to wholesaler in time
b
t
t for the Fulton Fish Market, New York (deflated)
N = population of Los Angeles SCSA in time t
T = variable "time" accounting for supply shifters not
otherwise observable given the available data.
All prices were deflated by the W.P.I, for "Farm products, processed foods and feeds", U.S. Bureau of Labor Statistics.
The results of the analysis of this general specification are
presented below in the "per capita", as opposed to "total", form.
This decision was based on the fact that, for all practical purposes,
the two specifications just mentioned, yield identical results.
The results of the first stage regression are:
P^1 = . 0094 + . 00439Prt
+ . 000537PrmS - . 000157D
W
t
(3.19) (4.72) Wt-1
(.742)
t-1
(-1.17)
+ . 0000709D
(.474)
+ . 000138D
, (1. 103)
+ . 0000736T - . 000018Dirnt
(3. 31)
(-.747) *
+ . 244P?Virgt - . 00595PrpSS - . 0000000002Y/N + . l07TP™ny
t
(2.57) t
(-3.40) Wt-1
(-3.34)
(.897) t-1
R2 = .919
F-statistic = 25. 891; 11, 25d.f.
81
In the second stage estimation of the demand equation,
P
D
t
is included as an explanatory variable and yields:
_wt
-~ = -.00406 - .00179P
- . 000453P
- . 0000958D
N
t
(-2.10) (-.695)Wt-l
(-.523) Wt-1
(-.578)
+ . 0000933D
(.471)
- . 000159D
(-.949)
+ . 0000000135Dimt + . 0894P^irgt
(.039)
*
(-.419) t
+ .0461PP
+ . 000000000076Y/N + . 5506P^
W
t
(1.36) t-1
(3.43)
(1. 21) t
R2 = . 573967
F-statistic = 3. 368; 10, 25 d-f.
Dur bin-Watson Statistic = 1. 693
where the t-values are in parenthesis below the coefficients.
The
t-values are statistically significant at the 90% level with 25 degrees
of freedom if they exceed the cr itical value 1.316, using a s ingletailed test.
The period of analysis is May 1972-December 1975, inclusive.
82
APPENDIX C
As explained in the chapter on model development, the simultaneous equation's specification was quantitatively evaluated in two
alternative forms depending upon whether the income variable was
deflated by the C.P.I, or not.
non-deflated specification.
Appendix B contains the results of the
The deflated personal disposable income
specification appears below.
Once again the model has been evaluated
in both "per capita" and "total" or aggregate terms.
The results are
not sufficiently different to warrant inclusion of both.
The "per
capita" results are given below.
In the second stage
*wt
P
t
is included as an explanatory
variable and yields'.
~wt
Q,
-~- = .0256 + . 00922P
+.00113PrmS - . 000342D
t
(4.06) (2.29) Wt-1
(1.34) Wt-1
(-2.08)
+ .0001 ID
(.654)
+ . 00038D
(1.68)
- .0000515Dimt + .7179P^rgt - . 0969PrpSS
t
(-1.53)
(2.33) t
(-2.69)Wt-l
- . 0000000631Y*/N - 2. IZIBP^
t
(-3.75)
(-2. 68) t
R2 = . 578107
F-statistic = 3.4257-, 10, 25
Durbin-Watson Statistic = 1.674
83
where the t-values are in parenthesis below the coefficients.
The
critical t-value is 1. 316 at the 90% level with 25 degrees of freedom,
noting once again the use of a single-tailed test.
In the first stage the regression yielded the following:
P^1 =. 01196 + .00337Prt
+ . 000863PrmS - . 000134D
W
t
(2.68)
(2.79) Wt-1
(.974)
t-1
(-.950)
+ .0000764D
(. 578)
+ . 0001 ID
(.708)
- .0000089T - . 0000171Dirnt + .3193P^irgt
(-1. 16)
(-. 656)
(3.45) t
- . 0486PrpSS + . 1253P^ny - . 00000003 16Y*/N
t
(-2. 67)wt-l
(. 960) t-1
(-2.71)
R2 = . 9098
F-statistic = 22.926; 11, 25 d.f.
where the t-values are in parenthesis below the coefficients.
The
t-values are statistically significant at the 90% level with 25 degrees
of freedom if they exceed the critical value 1. 316.
The period of
analysis is May, 1972-December, 1975, inclusive.
Note that Y*/N
is the deflated per capita personal disposable income for time
L.A. SCSA.
t,
84
APPENDIX D
For the third alternative specification being considered, namely
the price of rainbow trout dependent single equation model, ordinary
least squares (OLS) technique has been employed to estimate the
coefficients.
P
t
The results are as follows;
= . 00242 + . 00622P
+ . 0036P
- . 0000252D2 - . 000174D3
(1.11)
(5.51) Wt-1
(.587)Wt-l
(-.132)
(-1.00)
+ . 0000852D
(.481)
- . 06765PrpSS + . 22803Q'7t/N - . 000369Y/N
t
t
(-3.22) Wt-1
(1.28)
(-1.38)
R2 = . 770
F-statistic = 14. 29; 8, 34 d.f.
Durbin-Watson Statistic = 1.5 99
where the t-values are in parenthesis below the coefficients.
The
critical t-value at the 90% confidence level with 32 degrees of freedom
is 1.311.
The period of analysis is May 1972-December 1975, inclusive.